Methods for Analysis of Naturalistic Driving Data in Driver Behavior Research
Doctoral thesis, 2016
In the last several years, the focus of traffic safety research—especially when performed in association with the automotive industry—has shifted from preventing injury during a crash to avoiding the crash altogether or mitigating its effects. Pre-crash safety measures include intelligent safety systems (e.g., different levels of automated driving), infrastructure design, behavior-based safety, and policy-making. Understanding driver behavior is crucial in the development and evaluation of such measures. Naturalistic driving data (NDD) can facilitate this understanding by providing information about crash causation and contribute to the evaluation of pre-crash safety measures and the effects of driver behavior on safety. However, NDD’s complexity calls for new and better methods to fully exploit its advantages.
This thesis, together with the five included papers, addresses several gaps in current scientific knowledge by presenting novel methods for analyzing NDD that address multiple aspects of the development process for pre-crash safety measures. The chunking method (Paper I) helps to identify and overcome common biases in analysis of everyday-driving time-series data, while the expert-assessment-based crash-causation analysis method (Paper II, supported by Paper III) is a novel approach to studying crash causation through the analysis of NDD with video. Product and prototype development can be improved by utilizing counterfactual simulations, for which the choice of driver behavior model is shown to be crucial (Paper IV)—an awareness that was previously lacking. Being able to compare the effects of drivers’ specific behaviors (e.g., driver-vehicle interactions or in-vehicle secondary tasks) on safety could both speed up development of safety measures and improve vehicle designs and design guidelines. Methods to perform such comparisons through the combination of counterfactual glance behavior and pre-crash kinematics have been missing (they are provided in Paper V). This thesis further improves the evaluation of pre-crash safety measures by providing more robust analyses of everyday driving data (Paper I) and by demonstrating the importance of good mathematical models of driver behavior in virtual evaluation (Paper IV).
In summary, these new methods fill important research gaps and have the potential to improve the design of pre-crash safety measures through the use of NDD. Using NDD can augment our understanding of driver behavior and crash causation, important aspects of improving traffic safety and fulfilling Sweden’s Vision Zero.
safety benefit evaluation
crash causation
safety measures
automated driving
analysis
naturalistic driving data
driver behavior
counterfactual simulations
Room Delta, House Svea, Forskningsgången 4, Gothenburg
Opponent: Adjunct Professor Jim Sayer, Department of Civil and Environmental Engineering, University of Michigan, USA
Author
Jonas Bärgman
Chalmers, Applied Mechanics, Vehicle Safety
Chunking: a procedure to improve naturalistic data analysis
Accident Analysis and Prevention,;Vol. 58(2013)p. 309-317
Journal article
Using manual measurements on event recorder video and image processing algorithms to extract optical parameters.
Proceedings of the Seventh International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design,;(2013)p. 177-183
Paper in proceeding
How does glance behavior influence crash and injury risk? A ‘what-if’ counterfactual simulation using crashes and near-crashes from SHRP2
Transportation Research Part F: Traffic Psychology and Behaviour,;Vol. 35(2015)p. 152-169
Journal article
Analysis of the role of inattention in road crashes based on naturalistic on-board safety monitoring data
PROCEEDINGS of the 3rd International Conference on Driver Distraction and Inattention,;(2013)
Paper in proceeding
Bärgman, J., Boda, C.N., & Dozza, M., Counterfactual simulations applied to SHRP2 crashes: The effect of driver behavior models on safety benefit estimations of intelligent safety systems
Åtgärder för att minska antalet skadade och dödade i trafikolyckor inkluderar utformning av vägar, förarträning, lagstiftning samt nya säkerhetssystem i bil. Under de senaste årtiondena har fokus inom trafiksäkerhetsforskningen, särskilt för säkerhetssystem i bil, skiftat från system som avser att förhindra skador när väl en olycka har inträffat, till system som agerar innan en olycka, och då försöker undvika olyckan helt. Då en mycket stor andel av dagens trafikolyckor beror på förarbeteende är forskning för att förstå förares beteende och samspel med omgivningen och trafiksäkerhetsåtgärder avgörande för att utveckla effektiva åtgärder.
Denna avhandling möter behovet av nya metoder för att förstå förares beteende i relation till trafiksäkerhet. Avhandlingen kretsar kring användandet av naturalistisk kördata (NKD). NKD är data som samlas in i bilar som förarna använder i sin vardag under lång tid (månader till år). Typisk NKD omfattar video på föraren och omgivningen samt information om förarens samspel med bilen (t.ex. inbromsning, styrning och fordonets hastighet) och omgivningen (t.ex. avstånd till andra bilar).
Avhandlingen presenterar fem nya metoder som använder NKD och som på olika sätt syftar till att stötta utvecklingen av nya trafiksäkerhetsåtgärder som avser att undvika olyckor, med fokus på utveckling av nya säkerhetssystem i bil. Metoderna spänner från kvalitativ analys av faktiska olyckor på video för att förstå orsakssamband, till en metod för att virtuellt (genom datorsimulering) utvärdera vilken effekt förares blickbeteende har på trafiksäkerheten. I avhandlingen diskuteras också tre generella områden rörande analys av NKD med avseende på a) hur nästanolyckor kan användas som alternativ till faktiska olyckor i trafiksäkerhetsforskning, b) hur slutledning av orsakssamband kan göras med NKD, och c) hur NKD som samlas in av vinstdrivande företag kan användas för forskning.
Denna avhandling möter behovet av nya metoder för att förstå förares beteende i relation till trafiksäkerhet. Avhandlingen kretsar kring användandet av naturalistisk kördata (NKD). NKD är data som samlas in i bilar som förarna använder i sin vardag under lång tid (månader till år). Typisk NKD omfattar video på föraren och omgivningen samt information om förarens samspel med bilen (t.ex. inbromsning, styrning och fordonets hastighet) och omgivningen (t.ex. avstånd till andra bilar).
Avhandlingen presenterar fem nya metoder som använder NKD och som på olika sätt syftar till att stötta utvecklingen av nya trafiksäkerhetsåtgärder som avser att undvika olyckor, med fokus på utveckling av nya säkerhetssystem i bil. Metoderna spänner från kvalitativ analys av faktiska olyckor på video för att förstå orsakssamband, till en metod för att virtuellt (genom datorsimulering) utvärdera vilken effekt förares blickbeteende har på trafiksäkerheten. I avhandlingen diskuteras också tre generella områden rörande analys av NKD med avseende på a) hur nästanolyckor kan användas som alternativ till faktiska olyckor i trafiksäkerhetsforskning, b) hur slutledning av orsakssamband kan göras med NKD, och c) hur NKD som samlas in av vinstdrivande företag kan användas för forskning.
There are several ways to reduce the number of injuries and fatalities on our roads, including better infrastructure design, driver training, and legislation, and new in-vehicle safety systems. During the last decades, research on in-vehicle safety systems has shifted from systems mitigating injuries when the crash has happened to systems acting before the crash to avoid the crash altogether. One example of such a system, forward collision warning (FCW), warns the driver that a vehicle ahead is rapidly getting closer. Systems such as FCW interact with the driver, and driver behavior is a factor in most crashes, so it is important to understand how drivers interact with their surroundings and in-vehicle safety systems.
This thesis seeks to increase our understanding of drivers’ behaviors, and how they affect traffic safety, by presenting new methods for analyzing naturalistic driving data (NDD). This type of data is collected unobtrusively for months (or years) in vehicles that the participating drivers use in their everyday lives. Typical NDD include video of the driver and the surrounding traffic and information about driver-vehicle interactions (e.g., braking, steering, and vehicle speed) and the surroundings (e.g., distance to other vehicles).
The five new NDD analysis methods presented in the thesis aim to help increase traffic safety and save lives, especially by supporting the design of in-vehicle safety systems. The methods range from an expert-assessment-based method to understand why crashes happen (based on the review of video of actual crashes) to a method that uses virtual (computer) simulations to estimate the risks associated with different tasks that drivers perform while driving (e.g., tuning a radio or writing an SMS). In addition, three general topics about NDD analysis are discussed: (a) how (and if) near-crashes can be used instead of crashes in traffic safety analysis, (b) if it is possible to infer causation from observational data, and (c) the use of commercially collecte
This thesis seeks to increase our understanding of drivers’ behaviors, and how they affect traffic safety, by presenting new methods for analyzing naturalistic driving data (NDD). This type of data is collected unobtrusively for months (or years) in vehicles that the participating drivers use in their everyday lives. Typical NDD include video of the driver and the surrounding traffic and information about driver-vehicle interactions (e.g., braking, steering, and vehicle speed) and the surroundings (e.g., distance to other vehicles).
The five new NDD analysis methods presented in the thesis aim to help increase traffic safety and save lives, especially by supporting the design of in-vehicle safety systems. The methods range from an expert-assessment-based method to understand why crashes happen (based on the review of video of actual crashes) to a method that uses virtual (computer) simulations to estimate the risks associated with different tasks that drivers perform while driving (e.g., tuning a radio or writing an SMS). In addition, three general topics about NDD analysis are discussed: (a) how (and if) near-crashes can be used instead of crashes in traffic safety analysis, (b) if it is possible to infer causation from observational data, and (c) the use of commercially collecte
Subject Categories
Other Computer and Information Science
Transport Systems and Logistics
Applied Psychology
Human Computer Interaction
Vehicle Engineering
Robotics
Signal Processing
ISBN
978-91-7597-501-6
Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 4182
Publisher
Chalmers
Room Delta, House Svea, Forskningsgången 4, Gothenburg
Opponent: Adjunct Professor Jim Sayer, Department of Civil and Environmental Engineering, University of Michigan, USA